EVALUATION OF GROUND HYPERSPECTRAL REMOTE SENSING FOR PRECISE SATELLITE IMAGERY BASED LAND COVER CLASSIFICATION AND FORECASTING OF RICE GRAIN PRODUCTION

摘要

Growing population and increasing food demands necessitate the early and accurate estimates of grain production from crop based to now variety based. For which the knowledge of potential discriminates for the rice varieties during satellite imagery classification has become important prior to take satellite imagery. A field based experimental study was conducted in which pure stands of eight rice varieties were raised to maximum tillering stage. Hyper-spectral and multi-angular reflectance was took from the canopy of rice stands using a portable spectrometer. One nadir and two off nadir viewing angles and wavelength range from 400 nm to 1050 nm were considered for reflectance measurements. Reflectance obtained from the rice varieties were compared for specified viewing angles, wavelength bands and directions. Wavelength range and viewing direction which can potentially differentiated almost all varieties from each other were evaluated as 540 nm to 560 nm and perpendicular plane at considered viewing angles.